practice
If Google were created from scratch today, much of it
would be learned, not coded.

—Jeff Dean, Google Senior Fellow, Systems and Infrastructure Group

MACHINE LEARNING, OR ML, is all the rage today, and there are good reasons for that. Models created by machine-learning algorithms for problems such as spam filtering, speech and image recognition, language translation, and text understanding have many advantages over code written by human developers. Machine learning, however, is not as magical as it sounds at first. In fact, it is rather analogous to how human developers create code using test-driven development. 4 Given a training set of input-output pairs {(a,b)|a ∈ A, b∈B}, guess a function f ∈ A

→ B that passes all the given tests but also generalizes
to unspecified input values.
A big difference between human-written code
and learned models is that the latter are usually not
represented by text and hence are not understandable
by human developers or manipulable by existing tools.

The consequence is that none of the traditional software engineering techniques for conventional programs such as code reviews, source control, and debugging are applicable anymore. Since incomprehensibility is not unique to learned code, these aspects are not of concern here.

A more interesting divergence between machines and humans is that machines are less arrogant than humans, and they acknowledge uncertainty in their code by returning a probability distribution or confidence interval of possible answers f ∈ A → ℙ(B) instead of claiming to know the precise result for every input. For example, a learned image-recognition function

DOI: 10.1145/3052935

Article development led by queue.acm.org

Modern applications are increasingly using
probabilistic machine-learned models.
BY ERIK MEIJER
Making Money
Using Math